Functional Conservation of the Capacity for Ent-Kaurene Biosynthesis and an Associated Operon in Certain Rhizobia

Biochemistry, Biophysics and Molecular Biology Publications Biochemistry, Biophysics and Molecular Biology

1-2014

Functional Conservation of the Capacity for ent-Kaurene Biosynthesis and an Associated Operon in Certain Rhizobia

David M. Hershey Iowa State University

Xuan Lu Iowa State University, [email protected]

Jiachen Zi Iowa State University, [email protected]

Reuben J. Peters Iowa State University, [email protected]

Follow this and additional works at: https://lib.dr.iastate.edu/bbmb_ag_pubs

Part of the Biochemistry, Biophysics, and Structural Biology Commons, Genomics Commons, and the Plant Biology Commons

The complete bibliographic information for this item can be found at https://lib.dr.iastate.edu/ bbmb_ag_pubs/2. For information on how to cite this item, please visit http://lib.dr.iastate.edu/ howtocite.html.

This Article is brought to you for free and open access by the Biochemistry, Biophysics and Molecular Biology at Iowa State University Digital Repository. It has been accepted for inclusion in Biochemistry, Biophysics and Molecular Biology Publications by an authorized administrator of Iowa State University Digital Repository. For more information, please contact [email protected]. Functional Conservation of the Capacity for ent-Kaurene Biosynthesis and an Associated Operon in Certain Rhizobia

Abstract Bacterial interactions with plants are accompanied by complex signal exchange processes. Previously, the nitrogen-fixing symbiotic (rhizo)bacterium Bradyrhizobium japonicum was found to carry adjacent genes encoding two sequentially acting diterpene cyclases that together transform geranylgeranyl diphosphate to ent-kaurene, the olefin precursor to the gibberellin plant hormones. Species from the three other major genera of rhizobia were found to have homologous terpene synthase genes. Cloning and functional characterization of a representative set of these enzymes confirmed the capacity of each genus to produce ent-kaurene. Moreover, comparison of their genomic context revealed that these diterpene synthases are found in a conserved operon which includes an adjacent isoprenyl diphosphate synthase, shown here to produce the geranylgeranyl diphosphate precursor, providing a critical link to central metabolism. In addition, the rest of the operon consists of enzymatic genes that presumably lead to a more elaborated diterpenoid, although the production of gibberellins was not observed. Nevertheless, it has previously been shown that the operon is selectively expressed during nodulation, and the scattered distribution of the operon via independent horizontal gene transfer within the symbiotic plasmid or genomic island shown here suggests that such diterpenoid production may modulate the interaction of these particular symbionts with their host plants.

Keywords Biosynthetic pathways, genetics, diterpenes, mesorhizobium, synteny

Disciplines Biochemistry, Biophysics, and Structural Biology | Genomics | Plant Biology

Comments This article is from Journal of Bateriology 196 (2014): 100, doi:10.1128/JB.01031-13. Posted with permission.

This article is available at Iowa State University Digital Repository: https://lib.dr.iastate.edu/bbmb_ag_pubs/2 Functional Conservation of the Capacity for ent-Kaurene Biosynthesis and an Associated Operon in Certain Rhizobia

David M. Hershey,* Xuan Lu, Jiachen Zi, Reuben J. Peters ‹Department of Biochemistry, Biophysics & Molecular Biology, Iowa State University, Ames, Iowa, USA

Bacterial interactions with plants are accompanied by complex signal exchange processes. Previously, the nitrogen-fixing symbi- Downloaded from otic (rhizo)bacterium Bradyrhizobium japonicum was found to carry adjacent genes encoding two sequentially acting diterpene cyclases that together transform geranylgeranyl diphosphate to ent-kaurene, the olefin precursor to the gibberellin plant hormones. Species from the three other major genera of rhizobia were found to have homologous terpene synthase genes. Cloning and functional characterization of a representative set of these enzymes confirmed the capacity of each genus to produce ent- kaurene. Moreover, comparison of their genomic context revealed that these diterpene synthases are found in a conserved operon which includes an adjacent isoprenyl diphosphate synthase, shown here to produce the geranylgeranyl diphosphate precursor, providing a critical link to central metabolism. In addition, the rest of the operon consists of enzymatic genes that pre- http://jb.asm.org/ sumably lead to a more elaborated diterpenoid, although the production of gibberellins was not observed. Nevertheless, it has previously been shown that the operon is selectively expressed during nodulation, and the scattered distribution of the operon via independent horizontal gene transfer within the symbiotic plasmid or genomic island shown here suggests that such diterpenoid production may modulate the interaction of these particular symbionts with their host plants.

acteria play critical roles in biogeochemical cycles, such as the triguingly, the prevalence of insertion sequences and phage inte- Bfixation of nitrogen. Although nitrogen makes up approxi- gration is thought to promote rearrangement within these on December 11, 2014 by IOWA STATE UNIVERSITY mately 80% of the Earth’s atmosphere, its bioavailability remains symbiotic modules (15). a major limitation, e.g., to plant growth (1). This is due to the Previously, we characterized two terpene synthases from Bra- inability of plants to assimilate the diatomic nitrogen that occurs dyrhizobium japonicum USDA110 (16). These proved to be diter- naturally in the atmosphere (2). Among plants, legumes uniquely pene cyclases capable of successively converting the general diter- host bacteria from the Rhizobiales order of the Alphaproteobacteria penoid precursor (E,E,E)-geranylgeranyl diphosphate (GGPP) in nodules formed following invasion of their root cortical cells. into ent-copalyl diphosphate (ent-CPP) and, hence, to ent- Inside these nodules, the rhizobia develop into endosymbiont kaurene, a precursor to the gibberellin phytohormones (Fig. 1). bacteroids, fixing nitrogen in exchange for carbon from their plant The relevant genes, blr2149 and blr2150, then encode an ent- hosts (3). This agriculturally important collaboration is thought copalyl diphosphate synthase (CPS) and ent-kaurene synthase to be the main biological route for nitrogen fixation. Of particular (KS), respectively. Notably, these two genes fall into a more exten- relevance here, a number of the rhizobia have been shown to pro- sive operon that was originally defined by Tully et al. (17) and duce plant growth hormones, such as the gibberellins, which are suggested to be present in all rhizobia (18), although it did not thought to further promote growth of the host plant (4). appear to affect the ability of B. japonicum to nodulate soybean Both legume and rhizobial species exhibit a surprising amount (Glycine max) or fix nitrogen in the resulting nodules (19). In the of specificity with respect to symbiotic partners (5). Only rarely study described here, we investigated the functional conservation can a given rhizobial species nodulate more than a few closely of the CPS and KS from this operon and demonstrate the produc- related plants. This specificity is due to bacterial and plant factors tion of the upstream GGPP by the isoprenyl diphosphate synthase (2). The host plant secretes flavonoid inducers that elicit rhizobial encoded by the adjacent gene in the operon. In addition, we found production of lipochitooligosaccharide Nod factors, which are that this operon exhibits a scattered distribution within the rhizo- recognized by the host plant, with subsequent steps in the nodu- bia. While examples are found in all four major genera, with con- lation process being dependent on recognition of bacterial cell servation of the ability to produce ent-kaurene, the uneven distri- surface chemistry and effector proteins as well. However, the host bution of the operon suggests that such diterpenoid production plant also applies the usual defense mechanisms—e.g., microbe- provides a selective advantage only under certain conditions. associated molecular pattern-triggered immunity and R-gene recognition of bacterial effectors—to restrict nodulation by un- wanted strains (2, 3, 6). This complex signal exchange process Received 2 September 2013 Accepted 14 October 2013 exerts extreme evolutionary pressure on the rhizobia (7–9), which Published ahead of print 18 October 2013 can be inferred, in part, from the presence of large plasmids or Address correspondence to Reuben J. Peters, [email protected]. genomic islands with distinct GϩC contents relative to the GϩC * Present address: David M. Hershey, Department of Plant and Microbial Biology, content in the rest of the genome. These large plasmids or genomic University of California, Berkeley, Berkeley, California, USA. islands contain the large set of genes required for nodulation, in- D.M.H. and X.L. contributed equally to this article. cluding nitrogen fixation (5, 10–13). This presumably reflects the Copyright © 2014, American Society for Microbiology. All Rights Reserved. ability of horizontal transfer to spread these distinct genetic ele- doi:10.1128/JB.01031-13 ments, enabling nodulation by the recipient rhizobia (9, 14). In-

100 jb.asm.org Journal of Bacteriology p. 100–106 January 2014 Volume 196 Number 1 Rhizobial Diterpenoid Biosynthesis Operon

plasmids described below, along with pIRS (i.e., to increase the isoprenoid OPP OPP precursor pool, as described previously [23]). Liquid cultures (50 ml) of H the resulting recombinant E. coli strains were induced at an optical density CPS KS at 600 nm of 0.6, the pH was adjusted to 7.1, and the bacteria were grown H H at 16°C for 72 h. The cultures were then extracted with an equal volume of GGPP ent-CPP ent-kaurene hexanes. The organic extract was separated out and dried in a rotary ␮ FIG 1 Production of ent-kaurene by B. japonicum. Shown are the steps cata- evaporator, and the residue was resuspended in 100 l hexanes. This lyzed by the characterized ent-copalyl diphosphate synthase (CPS) and ent- concentrated extract was analyzed by gas chromatography (GC), carried kaurene synthase (ent-KS). out on a Varian (Palo Alto, CA) 3900 GC with a Saturn 2100 ion trap mass spectrometer (MS) in electron ionization (70 eV) mode. Samples (1 ␮l)

were injected in splitless mode at 50°C, and after holding for 3 min at Downloaded from 50°C, the oven temperature was raised at a rate of 14°C/min to 300°C, MATERIALS AND METHODS where it was held for an additional 3 min. MS data from m/z 90 to 600 were General. Unless otherwise noted, all molecular biology reagents were pur- collected starting 12 min after injection and were collected until the end of chased from Invitrogen and all other chemicals were from Fisher Scien- the run. The production of ent-kaurene was verified by comparison of the tific. B. japonicum USDA110 was obtained from Michael Sandowsky mass spectra and retention time to those of an authentic standard (enzy- (University of Minnesota), and Mesorhizobium loti MAFF303099, matically produced by the characterized CPS and KS from A. thaliana). Sinorhizobium fredii NGR234, and Rhizobium etli CFN42 were all ob- Mapping the diterpenoid biosynthesis operon. A 20-kb region sur- tained from Philip Poole (John Innes Centre), while Sinorhizobium meli- rounding each biochemically characterized KS was downloaded from the loti 1021 was obtained from Kathryn Jones (Florida State University). NCBI website and further analyzed. The predicted genes that either were http://jb.asm.org/ Escherichia coli was grown at 37 or 16°C on either NZY (for cloning) or TB homologous to those in the B. japonicum operon or had plausible pre- medium (for expression). Rhizobia were cultured with YEM medium at dicted functions in (di)terpenoid biosynthesis were identified by align- 28°C. When necessary, 1.8% agar was added to the relevant medium to ment and open reading frame prediction. In each case, the boundaries of pour plates. Where applicable, antibiotics were used at the following con- each operon were clear from the predicted functions of the adjacent genes centrations: chloramphenicol, 30 ␮g/ml; carbenicillin, 50 ␮g/ml; specti- (i.e., these have no plausible function in terpenoid biosynthesis). Putative nomycin, 50 ␮g/ml; and kanamycin, 50 ␮g/ml. Liquid cultures were RpoN and NifA binding sites were identified by searching for identical grown with vigorous shaking (200 rpm), generally in 250-ml Erlenmeyer matches in the upstream region of each operon to previously defined

flasks with 50 ml medium. Microanaerobic cultures were grown in YEM 16-nucleotide motifs (24). on December 11, 2014 by IOWA STATE UNIVERSITY medium with 10 mM KNO3 under an atmosphere of nitrogen gas (N2) Characterization of GGPS. Fragments from the 3= region of the and ϳ0.5% oxygen with moderate shaking (80 rpm) in rubber-stoppered operon from S. fredii, including genes for GGPS-CPS-KS or CPS-KS only, flasks, with the atmosphere exchanged every 12 h (N2 was blown into the were amplified from genomic DNA via PCR. These were cloned into flasks for 15 min). pZeroBlunt and then subcloned into a previously described S. meliloti Sequence retrieval and analysis. All sequences were retrieved from overexpression vector (25), pstb-LAFR5 (obtained from Kathryn Jones, the National Center for Biotechnology Information (NCBI) website. The Florida State University), using BamHI and EcoRI restriction sites intro- amino acid sequence of the previously characterized KS from B. japoni- duced by PCR, along with three upstream stop codons and an optimized cum (BjKS) was used as a query for BLAST searches against the Rhizobiales ribosome binding site. The resulting constructs were transformed into S. order (i.e., by restricting the search to this order, defined as taxid 356) on meliloti 1021 by triparental mating using E. coli strain MM294A carrying the NCBI website (20). This also was done using the amino acid sequence the construct and E. coli strain MT616 as the helper, as described previ- of CPS from B. japonicum (BjCPS) as the query sequence. Sequence anal- ously (25). These recombinant strains were grown for 5 days, and then the yses were carried out with the CLC Main Workbench program (version total culture was extracted with an equal volume of hexanes. This organic 6.8.4). Alignments used the following parameters: gap open cost, 10; gap extract was separated out and dried under a gentle stream of N2, with the extension cost, 10; and end gap cost, as any other. Trees were prepared residue then resuspended in 200 ␮l of hexanes for analysis by GC-MS, as using the neighbor-joining algorithm with a bootstrap analysis of 1,000 described above. replicates. PAUP analysis was used to confirm the topology of the result- Analysis of rhizobial diterpenoid production. Liquid cultures grown ing trees. The phylogenetic analyses whose results are presented here were under aerobic or microanaerobic conditions were harvested 3, 6, or 9 days carried out using genes encoding biochemically analogous proteins from after inoculation, and the cells were separated from the spent medium by a bacterial species as distantly related as possible as the designated out- centrifugation (15 min at 10,000 ϫ g). For analysis of the gibberellin group sequence. For CPS, this was from Streptomyces sp. strain KO-3988, content, the supernatant was acidified to pH 2.5 with acetic acid and then which falls within the Actinobacteria phylum, yet this Streptomyces sp. CPS extracted with an equal volume of ethyl acetate saturated with acetic acid (SsCPS) (GenBank accession number AB183750) also produces ent-CPP (1%, vol/vol). This organic extract was separated and passed over a 1-ml (21). For NifK, this was from Azotobacter vinelandii (AvNifK; GenBank HP-20 resin column, which was then eluted with 3 ml each of 1% acetic accession number Avin_01400), which falls within the Proteobacteria phy- acid in distilled H2O (dH2O) and 40% and 80% (vol/vol in dH2O with 1% lum but is in the distinct Gammaproteobacteria class. acetic acid) methanol. Each of these fractions was dried in a rotary evap- Cloning and characterization of CPS and KS. Genomic DNA was orator, and the residue was resuspended in 100 ␮l acetic acid-saturated isolated from rhizobia using a Generation capture kit (Qiagen) following ethyl acetate for analysis by GC-MS as described above. For analysis of the the manufacturer’s protocol. Each CPS and KS gene was amplified via ent-kaurene intermediate, the total culture was directly extracted with an PCR from genomic DNA using gene-specific primers and cloned into equal volume of hexanes, which was separated out and passed over a 1-ml pENTR-SD-dTOPO (Invitrogen). Biochemical characterization of the silica gel column to remove contaminating polar compounds. The result- CPS-KS pair from each species of rhizobia was carried out as described ing organic extract was dried under a gentle stream of N2, and the residue previously for B. japonicum (16). Briefly, the KS genes were subcloned into was resuspended in 100 ␮l of hexanes for analysis by GC-MS, again, as the plasmid pGG-DEST, which carries a plant GGPP synthase (GGPS), described above. and the CPS genes were subcloned into pDEST14. This enabled use of a previously described metabolic engineering system, which included constructs analogous to the CPS and KS from Arabidopsis thaliana (22). Ac- RESULTS cordingly, the E. coli strain OverExpress C41 (Lucigen) was transformed Identification of KS and CPS homologs in rhizobia. As noted with the various combinations of the pGG-DEST::CPS and pDEST14::KS above, the BjKS that directly produces ent-kaurene exhibits dis-

January 2014 Volume 196 Number 1 jb.asm.org 101 Hershey et al.

TABLE 1 Rhizobial KS and CPS homologs KS CPS GenBank % identity GenBank % identity Organism accession no. to BjKS accession no. to BjCPS Bradyrhizobium japonicum NP_768790 NP_768789 Bradyrhizobium elkanii WP_018270013 91 WP_016845990 92 Bradyrhizobium sp. strain WSM1253 WP_007600190 89 WP_007600189 91 Bradyrhizobium sp. strain WSM471 WP_007605894 88 WP_007605892 91 Mesorhizobium loti NP_106894 93 NP_106893 93 Mesorhizobium alhagi WP_008838313 94 WP_008838314 95 Downloaded from Mesorhizobium amorphae WP_006204703 93 WP_006204702 93 Mesorhizobium ciceri YP_004144785 92 YP_004144784 92 Mesorhizobium sp. strain STM 4661 WP_006329103 92 WP_006329109 93 Mesorhizobium sp. strain WSM4349 WP_018457688 92 WP_018457687 93 Sinorhizobium fredii NP_443948 92 NP_443949 95 Sinorhizobium meliloti WP_018098888 91 WP_018098887 94 Sinorhizobium medicae WP_018009727 90 WP_018009726 92 Rhizobium etli NP_659792 87 NP_659791 86 Rhizobium tropici YP_007335933 87 YP_007335932 90 http://jb.asm.org/ Rhizobium sp. strain CCGE 510 WP_007636919 88 WP_007636921 87 Rhizobium grahamii WP_016558477 71 WP_016558476 72 Rhizobium mesoamericanum WP_007539161 69 WP_007539159 72

tinct sequence homology relative to other characterized bacterial diterpene synthases. Accordingly, the BjKS sequence was used in on December 11, 2014 by IOWA STATE UNIVERSITY initial BLAST searches of the NCBI database to identify bacteria from the Rhizobiales order that contain homologous diterpene synthases. Notably, homologs were found in species from all four major genera of rhizobia, i.e., Rhizobium, Sinorhizobium, and Mesorhizobium, in addition to Bradyrhizobium. In each case, immediately upstream of the BjKS homolog was a homolog to BjCPS (Table 1) with 2-nucleotide-overlapping open reading frames, just like the 2-nucleotide-overlapping open reading frame found in B. japonicum. Interestingly, these were not conserved by bacterial phylogeny; e.g., the KS from the various species of Bradyrhi- zobium shared less sequence identity than BjKS and the KS from Mesorhizobium loti, which not only is in a distinct genus but also falls into the separate Phyllobacteriaceae family. Accordingly, KS and CPS appear to have been distributed via horizontal gene transfer. Consistent with such an inheritance mechanism, the KS gene is not found in all rhizobia (e.g., no homolog is present in Rhizobium leguminosarum, whose genome has been fully se- quenced [26], nor are homologs present in any of the genome sequences reported for various strains of S. meliloti, despite the presence of a homologous protein sequence annotated as being encoded by Sinorhizobium meliloti [27–29]). Functional characterization of representative CPSs and KSs. To investigate the ability of the CPS and KS homologs found in our bioinformatics search to cooperatively produce ent-kaurene, we analyzed these from one species from each genus, specifically, examples of species for which complete genome sequences have been reported, Mesorhizobium loti MAFF303099 (12), Sinorhizo- bium fredii NGR234 (30), and Rhizobium etli CFN42 (10). We cloned and characterized the CPS and KS from each of these species much as previously described for those from B. japonicum (16). Briefly, each pair of CPS and KS homologs was coexpressed FIG 2 Selected ion (m/z 272) chromatograms obtained by GC-MS demonstrating production of ent-kaurene from GGPP by coexpressing CPS and KS in recombinant E. coli also expressing a plant GGPP synthase from M. loti (A), S. fredii (B), and R. etli (C), along with an authentic standard (Abies grandis GGPS [AgGGPS]), which led to the production of (from coexpression of the CPS and KS from Arabidopsis thaliana)inE. coli kaurene (Fig. 2). To demonstrate stereospecificity, each CPS was (along with a GGPP synthase) (D).

102 jb.asm.org Journal of Bacteriology Rhizobial Diterpenoid Biosynthesis Operon

B. japonicum CYP112 CYP114 Fd SDR CYP117 GGPS CPS KS

S. fredii CYP112 CYP114 Fd SDR CYP117 GGPS CPS KS

M. loti CYP112 CYP114 Fd-SDR CYP117 GGPS CPS KS IDI MFS

R. etli CYP112 CYP114-Fd SDR CYP117 GGPSp CPS KS IDI GGPS2

FIG 3 Schematic of diterpenoid biosynthesis operon from the designated rhizobia (with the gene designations described in the text). Downloaded from expressed in recombinant E. coli with AgGGPS and the KS from cordingly, in R. etli and M. loti, accessory genes appear to have Arabidopsis thaliana (AtKS), which is speciﬁc for ent-CPP. In ad- been appended to the core diterpenoid biosynthesis operon dition, each KS was expressed in recombinant E. coli with AgGGPS (Fig. 3). and the ent-CPP-producing CPS from Arabidopsis thaliana Upon sequence comparisons of the core operon, that from R. (AtCPS). In each case, this led to the production of ent-kaurene etli appeared to be the most divergent, sharing Ͻ82% identity,

(data not shown), demonstrating a stereochemistry consistent while the others were Ն90% identical to each other. Even when http://jb.asm.org/ with that of the gibberellin plant hormones. These results con- excluding the compromised GGPSp, comparison of the other firmed a common catalytic activity for these distributed enzymatic genes from the R. etli operon revealed that these are only 86 to 89% genes and, importantly, that each genus identified here contains identical to those from the other rhizobia, which is less than the the capacity to produce ent-kaurene from GGPP. level of identity shared by the other rhizobia. Accordingly, the R. Defining a rhizobial diterpenoid biosynthetic operon. Given etli operon is clearly the most divergent, consistent with distribu- that the genes for BjCPS and BjKS are neighboring genes in what tion of the entire operon by horizontal gene transfer; e.g., despite has been proposed to be a more extensive operon (19), we exam- their common phylogenetic origin in the Rhizobiaceae family, R. ined the genomic context for each of the characterized CPSs and etli and S. fredii contain the most disparate rhizobial diterpenoid on December 11, 2014 by IOWA STATE UNIVERSITY KSs to determine if these were similarly set in a more broadly biosynthesis operons. conserved operon. Indeed, homologs to all of the other genes from Demonstrating production of GGPP and capacity for ent- the B. japonicum operon also were present, with retention of rel- kaurene production. Diterpenoid biosynthesis generally pro- ative gene order. In particular, homologs to the cytochromes P450 ceeds via the initial formation of a hydrocarbon skeletal structure, CYP112 and CYP114, a ferredoxin (Fd), a short-chain alcohol followed by oxidative elaboration (31). In the organization of the dehydrogenase/reductase (SDR), another cytochrome P450 rhizobial diterpenoid biosynthesis operons, it is notable that the (CYP117), an isoprenyl diphosphate synthase that presumably genes predicted to be involved in oxidation are in the 5= region, makes GGPP (GGPS), as well as the orthologous CPS and KS were with all those predicted to be involved in the formation of the detected. Although it should be noted that some of these genes cyclized olefin ent-kaurene falling in the 3= region. This includes were fused together in certain cases, i.e., the CYP114 and Fd in R. the putative GGPP synthase (GGPS), as bacteria do not necessarily etli and the Fd and SDR in M. loti, these still exhibited clear ho- produce GGPP, leading to the presence of a GGPS in all of the mology to the separate genes found elsewhere. Thus, these genes identified bacterial diterpenoid biosynthetic gene clusters (21, 32– define a core diterpenoid biosynthetic gene cluster/operon that is 37). The observed organization of the rhizobial diterpenoid bio- conserved across all four of the major rhizobial genera, sharing 80 synthetic operon suggests that the 3= and 5= regions might form to 92% nucleotide sequence identity. nominally independent subclusters, although no such subclusters Notably, the GGPS gene in R. etli appears to be disrupted. appear in the currently available sequence information. Neverthe- While some homologous sequence is present, there is a large less, we took advantage of this gene arrangement to demonstrate internal deletion, resulting in a clearly compromised open both the production of GGPP by the isoprenyl diphosphate syn- reading frame (accordingly, we suggest that this is a pseudo- thase and, hence, the ability of the operon to lead to the produc- gene and refer to it as GGPSp). However, R. etli contains an- tion of at least ent-kaurene. In particular, while initial efforts were other isoprenyl diphosphate synthase in close proximity to its directed at heterologous expression of the putative GGPP syn- core operon. Although this is not closely related to the GGPS thase in E. coli for use in coexpression studies such as those de- found within the operon and is in the opposite orientation, we scribed above, that ultimately proved unsuccessful. We then hypothesize that this might serve the same function (and refer turned to recombinant expression in a more closely related bacte- to it here as GGPS2). There is an intervening gene. However, rium, specifically, the 1021 strain of Sinorhizobium meliloti, whose this gene appears to encode an isopentenyl diphosphate reported genome sequence does not contain the rhizobial diter- isomerase (IDI), which balances the isoprenoid precursor sup- penoid biosynthesis operon (27). Accordingly, we overexpressed ply and, thus, similarly has a plausible role in (di)terpenoid the 3= region of the operon from the closely related S. fredii, either biosynthesis as well. Further analysis demonstrated that a ho- a fragment containing GGPS-CPS-KS or a fragment containing mologous IDI gene also occurs at the same position (3= to the CPS-KS only. Consistent with the usual lack of GGPP production KS) in M. loti. Intriguingly, M. loti further has a gene encoding in bacteria, expression of the CPS-KS genes alone in S. meliloti a major facilitator superfamily (MFS) member immediately 1021 did not result in the production of ent-kaurene, while expres- downstream of its IDI, and we speculate that this might be sion of the GGPS-CPS-KS genes did (Fig. 4). These results, then, involved in secretion of the diterpenoid natural product. Ac- confirm that the associated isoprenyl diphosphate synthase pro-

January 2014 Volume 196 Number 1 jb.asm.org 103 Hershey et al.

sis of the gibberellin phytohormones in both plants and fungi (40), coupled to previous reports of bacterial production of gibberellins (4), we have hypothesized that the capacity to produce this diterpene is indicative of the ability to produce gibberellins (16). However, although it has been reported that B. japonicum

produces gibberellin A3 (GA3) under standard liquid culture conditions (41), we have been unable to detect this or ent-kaurene from B. japonicum grown under the previously described conditions or even under microanaerobic conditions. In addition, we

were unable to detect production of ent-kaurene from any of the Downloaded from FIG 4 Selected ion (m/z 272) chromatograms obtained by GC-MS demon- other rhizobia examined here under either aerobic or microan- strating production of ent-kaurene from S. meliloti 1021 expressing GGPS- aerobic culture conditions. CPS-KS, but not CPS-KS alone, from S. fredii NGR234 (as indicated). While the rhizobial symbiotic modules (plasmid or genomic island) are known to be distributed via horizontal gene transfer, the diterpenoid biosynthetic operon seems to be separately dis- duces GGPP, providing a critical link to central metabolism, and tributed, albeit via selective integration into the symbiotic mod- further demonstrate the capacity of the operon to confer the abil- ule. For example, it has been noted that this cytochrome P450-rich

ity to produce at least ent-kaurene. gene cluster exhibits a different GϩC content relative to that in the http://jb.asm.org/ Evidence for a putative role in symbiosis. In each of the rhi- rest of the symbiotic plasmid in S. fredii NGR234 (42). The phy- zobia examined here, the diterpenoid biosynthesis operon is lo- logenetic relationship of the CPS genes from the analyzed rhizo- cated in the symbiotic module (plasmid or genomic island), sug- bial diterpenoid biosynthesis operons, for which homologs are gesting that the resulting natural product may play a role in the evident in other bacteria, is distinct from that of the gene for the symbiotic relationship of these rhizobia with their host plants. nitrogenase subunit NifK (Fig. 5). This suggests that the diterpe- Consistent with such a role, it has been shown in B. japonicum that noid biosynthetic operon, which is not found in all such rhizobial the genes in the diterpenoid biosynthesis operon are expressed symbiotic modules in any case, has been independently incorpo- upon bacterial differentiation in the nodules to the nitrogen-fix- rated into these symbiotic modules. Although it has already been on December 11, 2014 by IOWA STATE UNIVERSITY ing bacteroid form and to a much lesser extent under microan- shown that polar disruption of the initial CYP112 gene in B. aerobic culture conditions which mimic those found in the nod- japonicum has no discernible effect on its ability to nodulate soy- ules (38). This expression has further been shown to be dependent bean or fix nitrogen (19), the resulting diterpenoid natural prod- on the symbiosis-specific sigma factor RpoN and associated tran- uct may provide some selective advantage in the symbiotic growth scription factor NifA (39). Similarly, expression of the genes from phase of the rhizobia where it has been incorporated. the diterpenoid biosynthesis operon in R. etli has also been shown to depend on RpoN and NifA and occur under similar conditions, DISCUSSION i.e., in nodules and under microanaerobic conditions (24). In ad- The results presented here demonstrate a scattered distribution of dition, sequences suggestive of regulation by the RpoN-NifA a diterpenoid biosynthetic operon within the rhizobia, with func- regulon also appear upstream of the diterpenoid biosynthesis tional conservation of at least the capacity for the production of operon in the other two genera. In particular, these are putative ent-kaurene (Fig. 2 and 4). Although the final diterpenoid end binding sites for both NifA and RpoN and, specifically, are the product remains unclear at this time, both plant and fungi pro- same sites identified upstream of the diterpenoid biosynthesis duce ent-kaurene en route to the production of gibberellins (40), operon in R. etli (24). Accordingly, it seems likely that the suggesting that this operon may also lead to such phytohormone diterpenoid biosynthesis operon in M. loti and S. fredii is also production. The location of this operon in the symbiotic module under the control of the RpoN-NifA regulon and expressed of the relevant rhizobia, along with its previously demonstrated during symbiosis. transcription in response to bacteroid differentiation in nodules, Given that ent-kaurene is the olefin intermediate in biosynthe- indicates a putative role for the resulting diterpenoid natural

FIG 5 Molecular phylogenetic analysis of genes for the characterized rhizobial CPS (A) and genes for the nitrogenase subunit NifK (B) from the same rhizobia. SsCPS and AvNifK are the designated outgroup sequences, as described in the text. SfCPS, S. fredii CPS; MlCPS, M. loti CPS; ReCPS, R. etli CPS; ReNifK, R. etli NifK; MlNifK, M. loti NifK; SfNifK, S. fredii NifK; BjNifK, B. japonicum NifK.

104 jb.asm.org Journal of Bacteriology Rhizobial Diterpenoid Biosynthesis Operon product in the symbiotic relationship of these rhizobia with their 10. Gonzalez V, Santamaria RI, Bustos P, Hernandez-Gonzalez I, Me- host plants. While the exact physiological function of this diterpe- drano-Soto A, Moreno-Hagelsieb G, Janga SC, Ramirez MA, Jimenez- noid remains unclear at this time as well, the scattered distribution Jacinto V, Collado-Vides J, Davila G. 2006. The partitioned Rhizobium etli genome: genetic and metabolic redundancy in seven interacting rep- of the operon, which appears to be a result of its apparently inde- licons. Proc. Natl. Acad. Sci. U. S. A. 103:3834–3839. http://dx.doi.org/10 pendent horizontal gene transfer between symbiotic modules, .1073/pnas.0508502103. suggests that it provides a selective advantage only under certain 11. Kaneko T, Nakamura Y, Sato S, Minamisawa K, Uchiumi T, Sasamoto conditions. Nevertheless, the striking conservation of this diterpe- S, Watanabe A, Idesawa K, Iriguchi M, Kawashima K, Kohara M, Matsumoto M, Shimpo S, Tsuruoka H, Wada T, Yamada M, Tabata S. noid biosynthetic operon hints at its importance. Intriguingly, all 2002. Complete genomic sequence of nitrogen-fixing symbiotic bacte- the characterized operons are from rhizobia associated with de- rium Bradyrhizobium japonicum USDA110. DNA Res. 9:189–197. http: terminate, rather than indeterminate, nodules. While nodule type //dx.doi.org/10.1093/dnares/9.6.189. Downloaded from is specified by the host plant species (43, 44), rhizobial specificity 12. Kaneko T, Nakamura Y, Sato S, Asamizu E, Kato T, Sasamoto S, for plant host species indirectly controls bacterial nodulation phe- Watanabe A, Idesawa K, Ishikawa A, Kawashima K, Kimura T, Kishida Y, Kiyokawa C, Kohara M, Matsumoto M, Matsuno A, Mochizuki Y, notypes. Accordingly, it is tempting to speculate that the diterpe- Nakayama S, Nakazaki N, Shimpo S, Sugimoto M, Takeuchi C, Yamada noid product of this operon specifically plays a role in rhizobial M, Tabata S. 2000. Complete genome structure of the nitrogen-fixing interactions within determinate nodules. While there are exam- symbiotic bacterium Mesorhizobium loti. DNA Res. 7:331–338. http://dx ples of the operon found in rhizobia usually isolated from plants .doi.org/10.1093/dnares/7.6.331. 13. Mavingui P, Flores M, Guo X, Davila G, Perret X, Broughton WJ, that form indeterminate nodules (e.g., S. meliloti), it is unclear if Palacios R. 2002. Dynamics of genome architecture in Rhizobium sp. these are complete. For example, while Mesorhizobium sp. strain strain NGR234. J. Bacteriol. 184:171–176. http://dx.doi.org/10.1128/JB http://jb.asm.org/ WSM4349 was isolated from Biserrula, which forms indetermi- .184.1.171-176.2002. nate nodules, its operon has lost the gene for CYP112 and contains 14. Sullivan JT, Ronson CW. 1998. Evolution of rhizobia by acquisition of a inactivating mutations in the SDR. Thus, the role of this operon 500-kb symbiosis island that integrates into a Phe-tRNA gene. Proc. Natl. Acad. Sci. U. S. A. 95:5145–5149. http://dx.doi.org/10.1073/pnas.95.9 and the resulting diterpenoid in determinate versus indeterminate .5145. nodules poses an interesting direction for future investigations. 15. Viprey V, Rosenthal A, Broughton WJ, Perret X. 2000. Genetic snapshots of the Rhizobium species NGR234 genome. Genome Biol. 1: ACKNOWLEDGMENTS RESEARCH0014. http://dx.doi.org/10.1186/gb-2000-1-6-research0014. 16. Morrone D, Chambers J, Lowry L, Kim G, Anterola A, Bender K, Peters on December 11, 2014 by IOWA STATE UNIVERSITY We thank Philip Poole (John Innes Centre) for helpful discussion and RJ. 2009. Gibberellin biosynthesis in bacteria: separate ent-copalyl providing rhizobia and Kathryn Jones (Florida State University) for pro- diphosphate and ent-kaurene synthases in Bradyrhizobium japonicum. viding the pstb-LAFR5 expression vector. We also thank Matt Hillwig for FEBS Lett. 583:475–480. http://dx.doi.org/10.1016/j.febslet.2008.12.052. productive discussions. 17. Tully RE, van Berkum P, Lovins KW, Keister DL. 1998. Identifcation This work was supported by a grant from the National Science Foun- and sequencing of a cytochrome P450 gene cluster from Bradyrhizobium dation (MCB0919735) to R.J.P. japonicum. Biochim. Biophys. Acta 1398:243–255. http://dx.doi.org/10 .1016/S0167-4781(98)00069-4. 18. Keister DL, Tully RE, Van Berkum P. 1999. A cytochrome P450 gene REFERENCES cluster in the Rhizobiaceae. J. Gen. Appl. Microbiol. 45:301–303. http://dx 1. Vance CP. 2001. Symbiotic nitrogen fixation and phosphorus acquisition. .doi.org/10.2323/jgam.45.301. Plant nutrition in a world of declining renewable resources. Plant Physiol. 19. Tully RE, Keister DL. 1993. Cloning and mutagenesis of a cytochrome 127:390–397. http://dx.doi.org//10.1104/pp.010331. P-450 locus from Bradyrhizobium japonicum that is expressed anaerobi- 2. Den Herder G, Parniske M. 2009. The unbearable naivety of legumes in cally and symbiotically. Appl. Environ. Microbiol. 59:4136–4142. symbiosis. Curr. Opin. Plant Biol. 12:491–499. http://dx.doi.org/10.1016 20. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. 1990. Basic local /j.pbi.2009.05.010. alignment search tool. J. Mol. Biol. 215:403–410. 3. Jones KM, Kobayashi H, Davies BW, Taga ME, Walker GC. 2007. How 21. Kawasaki T, Kuzuyama T, Kuwamori Y, Matsuura N, Itoh N, Furihata rhizobial symbionts invade plants: the Sinorhizobium-Medicago model. K, Seto H, Dairi T. 2004. Presence of copalyl diphosphate synthase gene Nat. Rev. Microbiol. 5:619–633. http://dx.doi.org/10.1038/nrmicro1705. in an actinomycete possessing the mevalonate pathway. J. Antibiot. 57: 4. Bottini R, Cassan F, Piccoli P. 2004. Gibberellin production by bacteria 739–747. http://dx.doi.org/10.7164/antibiotics.57.739. and its involvement in plant growth promotion and yield increase. Appl. 22. Cyr A, Wilderman PR, Determan M, Peters RJ. 2007. A modular ap- Microbiol. Biotechnol. 65:497–503. http://dx.doi.org/10.1007/s00253 proach for facile biosynthesis of labdane-related diterpenes. J. Am. Chem. -004-1696-1. Soc. 129:6684–6685. http://dx.doi.org/10.1021/ja071158n. 5. Fauvart M, Michiels J. 2008. Rhizobial secreted proteins as determinants 23. Morrone D, Lowry L, Determan MK, Hershey DM, Xu M, Peters RJ. of host specificity in the rhizobium-legume symbiosis. FEMS Microbiol. 2010. Increasing diterpene yield with a modular metabolic engineering Lett. 285:1–9. http://dx.doi.org/10.1111/j.1574-6968.2008.01254.x. system in E. coli: comparison of MEV and MEP isoprenoid precursor 6. Yang S, Tang F, Gao M, Krishnan HB, Zhu H. 2010. R gene-controlled pathway engineering. Appl. Microbiol. Biotechnol. 85:1893–1906. http: host specificity in the legume-rhizobia symbiosis. Proc. Natl. Acad. Sci. //dx.doi.org/10.1007/s00253-009-2219-x. U. S. A. 107:18735–18740. http://dx.doi.org/10.1073/pnas.1011957107. 24. Salazar E, Diaz-Mejia JJ, Moreno-Hagelsieb G, Martinez-Batallar G, 7. Flores M, Morales L, Avila A, Gonzalez V, Bustos P, Garcia D, Mora Y, Mora Y, Mora J, Encarnacion S. 2010. Characterization of the NifA- Guo X, Collado-Vides J, Pinero D, Davila G, Mora J, Palacios R. 2005. RpoN regulon in Rhizobium etli in free life and in symbiosis with Phaseo- Diversification of DNA sequences in the symbiotic genome of Rhizobium lus vulgaris. Appl. Environ. Microbiol. 76:4510–4520. http://dx.doi.org etli. J. Bacteriol. 187:7185–7192. http://dx.doi.org/10.1128/JB.187.21 /10.1128/AEM.02007-09. .7185-7192.2005. 25. Jones KM. 2012. Increased production of the exopolysaccharide succi- 8. Gonzalez V, Acosta JL, Santamaria RI, Bustos P, Fernandez JL, Her- noglycan enhances Sinorhizobium meliloti 1021 symbiosis with the host nandez Gonzalez IL, Diaz R, Flores M, Palacios R, Mora J, Davila G. plant Medicago truncatula. J. Bacteriol. 194:4322–4331. http://dx.doi.org 2010. Conserved symbiotic plasmid DNA sequences in the multireplicon /10.1128/JB.00751-12. pangenomic structure of Rhizobium etli. Appl. Environ. Microbiol. 76: 26. Young JP, Crossman LC, Johnston AW, Thomson NR, Ghazoui ZF, 1604–1614. http://dx.doi.org/10.1128/AEM.02039-09. Hull KH, Wexler M, Curson AR, Todd JD, Poole PS, Mauchline TH, 9. Johnston AWB, Beynon JL, Buchanan-Wollaston AV, Setchell SM, East AK, Quail MA, Churcher C, Arrowsmith C, Cherevach I, Chill- Hirsch PR, Beringer JE. 1978. High frequency transfer of nodulating ingworth T, Clarke K, Cronin A, Davis P, Fraser A, Hance Z, Hauser H, ability between strains and species of Rhizobium. Nature 276:634–636. Jagels K, Moule S, Mungall K, Norbertczak H, Rabbinowitsch E, Sand- http://dx.doi.org/10.1038/276634a0. ers M, Simmonds M, Whitehead S, Parkhill J. 2006. The genome of

January 2014 Volume 196 Number 1 jb.asm.org 105 Hershey et al.

Rhizobium leguminosarum has recognizable core and accessory compo- of brasilicardin A, a unique diterpenoid. J. Antibiot. 61:164–174. http://dx nents. Genome Biol. 7:R34. http://dx.doi.org/10.1186/gb-2006-7-4-r34. .doi.org/10.1038/ja.2008.126. 27. Galibert F, Finan TM, Long SR, Puhler A, Abola P, Ampe F, Barloy-Hubler 35. Kim SY, Zhao P, Igarashi M, Sawa R, Tomita T, Nishiyama M, Kuzuy- F, Barnett MJ, Becker A, Boistard P, Bothe G, Boutry M, Bowser L, ama T. 2009. Cloning and heterologous expression of the cyclooctatin Buhrmester J, Cadieu E, Capela D, Chain P, Cowie A, Davis RW, Dreano biosynthetic gene cluster afford a diterpene cyclase and two p450 hydroxy- S, Federspiel NA, Fisher RF, Gloux S, Godrie T, Goffeau A, Golding B, lases. Chem. Biol. 16:736–743. http://dx.doi.org/10.1016/j.chembiol.2009 Gouzy J, Gurjal M, Hernandez-Lucas I, Hong A, Huizar L, Hyman RW, .06.007. Jones T, Kahn D, Kahn ML, Kalman S, Keating DH, Kiss E, Komp C, 36. Mann FM, Xu M, Davenport EK, Peters RJ. 2012. Functional charac- Lelaure V, Masuy D, Palm C, Peck MC, Pohl TM, Portetelle D, Purnelle B, terization and evolution of the isotuberculosinol operon in Mycobacte- Ramsperger U, Surzycki R, Thebault P, Vandenbol M, et al. 2001. The rium tuberculosis and related mycobacteria. Front. Microbiol. 3:368. http: composite genome of the legume symbiont Sinorhizobium meliloti. Science //dx.doi.org/10.3389/fmicb.2012.00368. 293:668–672. http://dx.doi.org/10.1126/science.1060966. 37. Smanski MJ, Yu Z, Casper J, Lin S, Peterson RM, Chen Y, Wendt- Downloaded from 28. Schneiker-Bekel S, Wibberg D, Bekel T, Blom J, Linke B, Neuweger H, Pienkowski E, Rajski SR, Shen B. 2011. Dedicated ent-kaurene and Stiens M, Vorholter FJ, Weidner S, Goesmann A, Puhler A, Schluter A. ent-atiserene synthases for platensimycin and platencin biosynthesis. 2011. The complete genome sequence of the dominant Sinorhizobium Proc. Natl. Acad. Sci. U. S. A. 108:13498–13503. http://dx.doi.org/10.1073 meliloti field isolate SM11 extends the S. meliloti pan-genome. J. Biotech- /pnas.1106919108. nol. 155:20–33. http://dx.doi.org/10.1016/j.jbiotec.2010.12.018. 38. Pessi G, Ahrens CA, Rehrauer H, Lindemann A, Hauser F, Fischer H-M, 29. Weidner S, Baumgarth B, Gottfert M, Jaenicke S, Puhler A, Schneiker- Hennecke H. 2007. Genome-wide transcript analysis of Bradyrhizobium japonicum bacteroids in soybean root nodules. Mol. Plant Microbe Interact. Bekel S, Serrania J, Szczepanowski R, Becker A. 2013. Genome sequence 20:1353–1363. http://dx.doi.org/10.1094/MPMI-20-11-1353. of Sinorhizobium meliloti Rm41. Genome Announc. 1(1):e00013–12. 39. Hauser F, Pessi G, Friberg M, Weber C, Rusca N, Lindemann A, Fischer http://dx.doi.org/10.1128/genomeA.00013-12. H-M, Hennecke H. 2007. Dissection of the Bradyrhizobium japonicum http://jb.asm.org/ 30. Schmeisser C, Liesegang H, Krysciak D, Bakkou N, Le Quere A, Wollherr NifAϩsigma54 regulon, and identification of a ferredoxin gene (fdxN) for A, Heinemeyer I, Morgenstern B, Pommerening-Roser A, Flores M, Pala- symbiotic nitrogen fixation. Mol. Genet. Genomics 278:255–271. http: cios R, Brenner S, Gottschalk G, Schmitz RA, Broughton WJ, Perret X, //dx.doi.org/10.1007/s00438-007-0246-9. Strittmatter AW, Streit WR. 2009. Rhizobium sp. strain NGR234 possesses a 40. Hedden P, Phillips AL, Rojas MC, Carrera E, Tudzynski B. 2001. remarkable number of secretion systems. Appl. Environ. Microbiol. 75:4035– Gibberellin biosynthesis in plants and fungi: a case of convergent evolu- 4045. http://dx.doi.org/10.1128/AEM.00515-09. tion? J. Plant Growth Regul. 20:319–331. http://dx.doi.org/10.1007 31. Peters RJ. 2010. Two rings in them all: the labdane-related diterpenoids. /s003440010037. Nat. Prod. Rep. 27:1521–1530. http://dx.doi.org/10.1039/c0np00019a. 41. Boiero L, Perrig D, Masciarelli O, Penna C, Cassán F, Luna V. 2007.

32. Hamano Y, Dairi T, Yamamoto M, Kawasaki T, Kaneda K, Kuzuyama Phytohormone production by three strains of Bradyrhizobium japonicum and on December 11, 2014 by IOWA STATE UNIVERSITY T, Itoh N, Seto H. 2001. Cloning of a gene cluster encoding enzymes possible physiological and technological implications. Appl. Microbiol. Bio- responsible for the mevalonate pathway from a terpenoid-antibiotic- technol. 74:874–880. http://dx.doi.org/10.1007/s00253-006-0731-9. producing Streptomyces strain. Biosci. Biotechnol. Biochem. 65:1627– 42. Freiberg C, Fellay R, Bairoch A, Broughton WJ, Rosenthal A, Perret X. 1635. http://dx.doi.org/10.1271/bbb.65.1627. 1997. Molecular basis of symbiosis between Rhizobium and legumes. Na- 33. Durr C, Schnell H-J, Luzhetskyy A, Murillo R, Weber M, Welzel K, ture 387:394–401. http://dx.doi.org/10.1038/387394a0. Vente A, Bechthold A. 2006. Biosynthesis of the terpene phenalinolac- 43. Oke V, Long SR. 1999. Bacteroid formation in the Rhizobium-legume tone in Streptomyces sp. Tu6071: analysis of the gene cluster and genera- symbiosis. Curr. Opin. Microbiol. 2:641–646. http://dx.doi.org/10.1016 tion of derivatives. Chem. Biol. 13:365–377. http://dx.doi.org/10.1016/j /S1369-5274(99)00035-1. .chembiol.2006.01.011. 44. Oono R, Schmitt I, Sprent JI, Denison RF. 2010. Multiple evolutionary 34. Hayashi Y, Matsuura N, Toshima H, Itoh N, Ishikawa J, Mikami Y, origins of legume traits leading to extreme rhizobial differentiation. New Dairi T. 2008. Cloning of the gene cluster responsible for the biosynthesis Phytol. 187:508–520. http://dx.doi.org/10.1111/j.1469-8137.2010.03261.x.

106 jb.asm.org Journal of Bacteriology